Sensor networks are commonly made of a very large number of nodes. In fact, for certain sensor networks, such as advanced metering infrastructure (AMI) networks, it is not uncommon to have tens of millions of meters with sensors. AMI network sensors may, for example, sense power levels, power consumption, and certain alarm triggering events (e.g., power loss/blackouts, brownouts, or spikes, etc.). Often, the devices having attached sensors (e.g., meters) are arranged in groups, where N devices would be affected by the same events, such as being connected to the same centralized links (e.g., a distribution power cable), being in a same location (e.g., for fire detection), etc. When such an event occurs, (e.g., one of these links/cables fail), a subset M of the devices are likely to be affected. Such failure triggers the generation of a set of alarms (an alarm storm) by the sensor network, leading to highly congested networks (wired/wireless) where alarms may be lost, or delayed (at best).